NO. 4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC 83 



In this figure are also given the yearly temperatures in Hamburg 

 for the interval September to August as a dotted curve and in 

 figure 43 for January to December as a dotted curve, the scale 

 being indicated at the right. As we should expect, the agreement 

 is here not so good. For purposes of comparison, we have given 

 in this figure also the curves for the temperature of the air in 

 northwestern England (N. W. England) ; for the stations around 

 the English Channel (E. Kanal) ; for Vliessingen and for Borkum. 

 We can detect in the figure a gradual transition in these several 

 curves. 



If it is true that an agreement in« the above mentioned sense 

 exists between the difference in the surface temperature of the mid- 

 dle and eastern side of the Atlantic Ocean and the surface tempera- 

 ture in the proximity to the Channel at Liepe's station i, then this 

 correspondence should appear by comparison of Petersen s material 

 with Liepe's material, even though Petersen's material, as we have 

 already said, is not particularly complete on account of too small 

 fields having been used. The uppermost full drawn curve of figure 

 43, curve I, shows the difference for February between the anomalies 

 of the mean temperature of Petersen's stations 5 and 6 and the 

 anomalies of the mean temperature of his stations i and 2. These 

 stations correspond to our two 10° fields 30° to 39° west longitude 

 and 10° to 19° west longitude. The scale of this curve is inverted. 

 At the upper right hand corner, covering the interval from 1898 to 

 1910, curve II, dotted, shows the difference : Surface temperature 

 30° to 39° west longitude minus the surface temperature 10° to 

 19° west longitude from all our fields combined. The full drawn 

 curve III in figure 43 gives the temperature anomalies in February 

 for Liepe's station i. The full drawn curve IV gives the air tem- 

 perature in Hamburg in February according to Thraen (1915). 

 Between these curves there is well marked correspondence, particu- 

 larly in the latter part after 1892, when as we should expect the 

 observations become more complete and trustworthy. The curves 

 for the mean temperatures for the year (September to August) 

 for the surface of the ocean at Liepe's station i and for the air in 

 Hamburg are also shown in figure 43 (III and IV, dotted). These 

 curves also show good correspondence although with more excep- 

 tions. Particularly the yearly temperatures September to August, 

 1903, at Hamburg is too low, and shows at this point very little cor- 

 respondence. The yearly temperature for the calendar year 1903 

 is on the other hand somewhat high. The average temperature of 

 the calendar years show, moreover, a marked minimum in the year 

 1902 (see the dotted line IV in figure 43). 



